Humans are naturally drawn to visual content, and even though video dominates online information consumption, images remain a crucial part of the digital experience. With over 80% of online traffic being video, the cost of data bandwidth is a major concern for companies like Google. Despite previous efforts with WebP, a failed attempt to surpass JPG, PNG, and GIF, Google has shifted its focus to improving JPEG with jpegli.
Jpegli is a JPEG coding library that includes both an encoder and decoder. What sets it apart is its adherence to the original JPEG standard and compatibility with existing decoders like browsers and image viewers. Utilizing adaptive quantization and psychovisual modeling, jpegli reduces noise and improves image quality, compressing images up to 35% more than traditional JPEG codecs while maintaining visual quality. This is a significant advancement in saving online bandwidth, especially for platforms like Google Photos.
With jpegli, images appear clearer with fewer artifacts, and the library offers support for images with 10+ bits per component without sacrificing compatibility. Despite its advanced features, jpegli is comparable in speed to other coding libraries, ensuring efficient processing without added computational resources.
A higher ELO score indicates a better aggregate performance
Google has made the jpegli source code available on GitHub, allowing interested individuals to explore and potentially utilize the library today.
